基于小波的钦州水稻产量多时间尺度分析与预测

利用小波诊断技术对广西钦州1950～2005年水稻产量和播种面积进行了多时间尺度分析，并利用基于小波的ARIMA模型进行了预测。分析结果表明：56年来，钦州水稻总产量和单位产量波动具有明显的3a、7a和25a特征时间尺度，播种面积7a特征时间尺度主要受农村土地制度改革影响。基于小波的ARIMA模型在水稻产量、播种面积预测方面精度很高，预测误差与气象灾害和土地政策变化有关。利用基于小波的水稻产量多时间尺度分析与预测方法，可以辅助水稻产量增减周期的分析以及对未来趋势的判断，对于结合供求关系合理调整种植面积，促进农业可持续发展提供帮助。

Multiple Time Scale Analysis and Forecast of Paddy Yield in Qinzhou Based on Wavelet Analysis

The changes in yield of paddy and sown area in Qinzhou from 1950 to 2005 have been analyzed at multi-time scales with Morlet wavelet technique. The results show that the yield per unit area and total yield of paddy have 3-year,7-year and 25-year characteristic time scales from the figures of wavelet coefficient. The 7-year characteristic time scale of sown area of paddy is influenced by the changes of rural land policies. The simulation results of the yield per unit area and total yield of paddy show the accuracy and efficiency of the wavelet analysis-based ARIMA model, and the prediction error is mainly caused by meteorological disasters and the changes of rural land policies. Multiple time scale analysis and forecast of paddy yield based on wavelet analysis can be used to analyze the periods of increasing yield and decreasing yield of paddy and indicate the trend of future yield. The model will help adjust the sown area combining with the relationship of supply and demand and promote sustainable agricultural development.